bug 398419 - leak processing scripts take the most time on Linux
recode the leak processing from shell to awk. r=slavo with fix for solaris (solaris requires nawk rather than awk) git-svn-id: svn://10.0.0.236/trunk@237369 18797224-902f-48f8-a5cc-f745e15eee43
This commit is contained in:
rrelyea%redhat.com
@@ -79,6 +79,7 @@ memleak_init()
|
||||
OLD_LIBRARY_PATH=${LD_LIBRARY_PATH}
|
||||
TMP_LIBDIR="${HOSTDIR}/tmp$$"
|
||||
TMP_STACKS="${HOSTDIR}/stacks$$"
|
||||
TMP_COUNT="${HOSTDIR}/count$$"
|
||||
|
||||
PORT=${PORT:-8443}
|
||||
|
||||
@@ -107,6 +108,7 @@ memleak_init()
|
||||
case "${OS_NAME}" in
|
||||
"SunOS")
|
||||
DBX=`which dbx`
|
||||
AWK=nawk
|
||||
|
||||
if [ $? -eq 0 ] ; then
|
||||
echo "${SCRIPTNAME}: DBX found: ${DBX}"
|
||||
@@ -140,6 +142,7 @@ memleak_init()
|
||||
;;
|
||||
"Linux")
|
||||
VALGRIND=`which valgrind`
|
||||
AWK=awk
|
||||
|
||||
if [ $? -eq 0 ] ; then
|
||||
echo "${SCRIPTNAME}: Valgrind found: ${VALGRIND}"
|
||||
@@ -179,6 +182,7 @@ memleak_init()
|
||||
|
||||
tbytes=0
|
||||
tblocks=0
|
||||
truns=0;
|
||||
|
||||
MEMLEAK_DBG=1
|
||||
export MEMLEAK_DBG
|
||||
@@ -518,138 +522,69 @@ parse_logfile_dbx()
|
||||
########################################################################
|
||||
parse_logfile_valgrind()
|
||||
{
|
||||
in_mel=0
|
||||
in_sum=0
|
||||
bin_name=""
|
||||
${AWK} '
|
||||
BEGIN {
|
||||
in_mel=0;
|
||||
in_sum=0;
|
||||
bytes=0;
|
||||
blocks=0;
|
||||
runs=0;
|
||||
stack_string="";
|
||||
bin_name=""; }
|
||||
!/==[0-9]*==/ {
|
||||
if ( $1 == "'${VALGRIND}'" )
|
||||
bin_name = $8 ;
|
||||
next;
|
||||
}
|
||||
/blocks are/ {
|
||||
in_mel = 1;
|
||||
stack_string="";
|
||||
next;
|
||||
}
|
||||
/LEAK SUMMARY/ {
|
||||
in_sum=1;
|
||||
runs += 1;
|
||||
next;
|
||||
}
|
||||
/^==[0-9]*== *$/ {
|
||||
if (in_mel)
|
||||
print bin_name stack_string;
|
||||
in_sum = 0;
|
||||
in_mel = 0;
|
||||
next;
|
||||
}
|
||||
in_mel == 1 {
|
||||
new_line = $4;
|
||||
if ( new_line == "(within")
|
||||
new_line = "*";
|
||||
stack_string = "/" new_line stack_string;
|
||||
}
|
||||
in_sum == 1 {
|
||||
for (i=2; i <= NF; i++) {
|
||||
if ($i == "bytes") {
|
||||
str = $(i-1);
|
||||
gsub(",","",str);
|
||||
bytes += str;
|
||||
}
|
||||
if ($i == "blocks.") {
|
||||
str = $(i-1);
|
||||
gsub(",","",str);
|
||||
blocks += str;
|
||||
}
|
||||
}
|
||||
}
|
||||
END {
|
||||
print "# " bytes " bytes " blocks " blocks in " runs " runs" > "/dev/stderr";
|
||||
}' 2> ${TMP_COUNT}
|
||||
|
||||
while read line
|
||||
do
|
||||
gline=`echo "${line}" | grep "^=="`
|
||||
if [ -z "${gline}" ] ; then
|
||||
gline=`echo "${line}" | grep "^${VALGRIND} " | sed "s:.*/::"`
|
||||
if [ -n "${gline}" ] ; then
|
||||
bin_name=`echo "${line}" | cut -d" " -f8`
|
||||
fi
|
||||
continue
|
||||
fi
|
||||
|
||||
line=`echo "${line}" | sed "s/==[0-9]*==\s*\(.*\)/\1/"`
|
||||
|
||||
gline=`echo "${line}" | grep "blocks are"`
|
||||
if [ -n "${gline}" ] ; then
|
||||
in_mel=1
|
||||
mel_line=0
|
||||
stack_string=""
|
||||
else
|
||||
gline=`echo "${line}" | grep "LEAK SUMMARY"`
|
||||
if [ -n "${gline}" ] ; then
|
||||
in_sum=1
|
||||
mel_line=0
|
||||
fi
|
||||
fi
|
||||
|
||||
if [ -z "${line}" ] ; then
|
||||
if [ ${in_mel} -eq 1 ] ; then
|
||||
in_mel=0
|
||||
echo "${bin_name}${stack_string}"
|
||||
elif [ ${in_sum} -eq 1 ] ; then
|
||||
in_sum=0
|
||||
fi
|
||||
fi
|
||||
|
||||
if [ ${in_mel} -eq 1 ] ; then
|
||||
mel_line=`expr ${mel_line} + 1`
|
||||
|
||||
if [ ${mel_line} -ge 2 ] ; then
|
||||
new_line=`echo "${line}" | sed "s/[^:]*:\ \(\S*\).*/\1/"`
|
||||
if [ "${new_line}" = "(within" ] ; then
|
||||
new_line="*"
|
||||
fi
|
||||
stack_string="/${new_line}${stack_string}"
|
||||
fi
|
||||
elif [ ${in_sum} -eq 1 ] ; then
|
||||
mel_line=`expr ${mel_line} + 1`
|
||||
|
||||
if [ ${mel_line} -ge 2 ] ; then
|
||||
gline=`echo "${line}" | grep "bytes.*blocks"`
|
||||
if [ -n "${gline}" ] ; then
|
||||
lbytes=`echo "${line}" | sed "s/.*: \(.*\) bytes.*/\1/" | sed "s/,//g"`
|
||||
lblocks=`echo "${line}" | sed "s/.*bytes in \(.*\) blocks.*/\1/" | sed "s/,//g"`
|
||||
|
||||
tbytes=`expr "${tbytes}" + "${lbytes}"`
|
||||
tblocks=`expr "${tblocks}" + "${lblocks}"`
|
||||
else
|
||||
in_sum=0
|
||||
fi
|
||||
fi
|
||||
fi
|
||||
done
|
||||
}
|
||||
|
||||
############################# log_compare ##############################
|
||||
# local shell function to check if selected stack is found in the list
|
||||
# of ignored stacks
|
||||
########################################################################
|
||||
log_compare()
|
||||
{
|
||||
BUG_ID=""
|
||||
|
||||
while read line
|
||||
do
|
||||
LINE="${line}"
|
||||
STACK="${stack}"
|
||||
|
||||
if [ "${LINE}" = "${STACK}" ] ; then
|
||||
return 0
|
||||
fi
|
||||
|
||||
NEXT=0
|
||||
|
||||
gline=`echo "${LINE}" | grep '^#'`
|
||||
if [ -n "${gline}" ] ; then
|
||||
BUG_ID="${LINE}"
|
||||
NEXT=1
|
||||
fi
|
||||
|
||||
gline=`echo "${LINE}" | grep '*'`
|
||||
if [ -z "${gline}" ] ; then
|
||||
NEXT=1
|
||||
fi
|
||||
|
||||
while [ "${LINE}" != "" -a ${NEXT} -ne 1 ]
|
||||
do
|
||||
L_WORD=`echo "${LINE}" | cut -d '/' -f1`
|
||||
gline=`echo "${LINE}" | grep '/'`
|
||||
if [ -n "${gline}" ] ; then
|
||||
LINE=`echo "${LINE}" | cut -d '/' -f2-`
|
||||
else
|
||||
LINE=""
|
||||
fi
|
||||
|
||||
S_WORD=`echo "${STACK}" | cut -d '/' -f1`
|
||||
gline=`echo "${STACK}" | grep '/'`
|
||||
if [ -n "${gline}" ] ; then
|
||||
STACK=`echo "${STACK}" | cut -d '/' -f2-`
|
||||
else
|
||||
STACK=""
|
||||
fi
|
||||
|
||||
if [ "${L_WORD}" = "**" ] ; then
|
||||
return 0
|
||||
fi
|
||||
|
||||
if [ "${L_WORD}" != "${S_WORD}" -a "${L_WORD}" != "*" -a "${S_WORD}" != "*" ] ; then
|
||||
NEXT=1
|
||||
break
|
||||
fi
|
||||
done
|
||||
|
||||
if [ "${LINE}" = "" -a "${STACK}" = "" ] ; then
|
||||
return 0
|
||||
fi
|
||||
done
|
||||
|
||||
return 1
|
||||
# sigh it would be nice to just pipe stderr and let stdout go by I've never been
|
||||
# able to convince any shell to do that correctly, so we are reduced to using a temp
|
||||
# file
|
||||
read hash lbytes bytes_str lblocks blocks_str in_str lruns rest < ${TMP_COUNT}
|
||||
tbytes=`expr "${tbytes}" + "${lbytes}"`
|
||||
tblocks=`expr "${tblocks}" + "${lblocks}"`
|
||||
truns=`expr "${truns}" + "${lruns}"`
|
||||
rm ${TMP_COUNT}
|
||||
}
|
||||
|
||||
############################# check_ignored ############################
|
||||
@@ -657,24 +592,83 @@ log_compare()
|
||||
########################################################################
|
||||
check_ignored()
|
||||
{
|
||||
ret=0
|
||||
|
||||
while read stack
|
||||
do
|
||||
log_compare < ${IGNORED_STACKS}
|
||||
if [ $? -eq 0 ] ; then
|
||||
if [ ${BUG_ID} != "" ] ; then
|
||||
echo "IGNORED STACK (${BUG_ID}): ${stack}"
|
||||
else
|
||||
echo "IGNORED STACK: ${stack}"
|
||||
fi
|
||||
else
|
||||
ret=1
|
||||
echo "NEW STACK: ${stack}"
|
||||
fi
|
||||
done
|
||||
|
||||
return ${ret}
|
||||
${AWK} -F/ '
|
||||
BEGIN {
|
||||
ignore="'${IGNORED_STACKS}'"
|
||||
# read in the ignore file
|
||||
BUGNUM="";
|
||||
count = 0;
|
||||
new = 0;
|
||||
while ((getline line < ignore) > 0) {
|
||||
if (line ~ "^#[0-9]+") {
|
||||
BUGNUM = line;
|
||||
} else if (line ~ "^#") {
|
||||
continue;
|
||||
} else if (line == "") {
|
||||
continue;
|
||||
} else {
|
||||
bugnum_array[count] = BUGNUM;
|
||||
line_array[count] = line;
|
||||
depth_array[count]= split(line, tmp_array, "/");
|
||||
for (i=1; i <= depth_array[count]; i++ ) {
|
||||
ignore_array[count,i] = tmp_array[i];
|
||||
}
|
||||
count++;
|
||||
}
|
||||
}
|
||||
}
|
||||
{
|
||||
match_found = 0;
|
||||
for (i=0 ; i < count; i++) {
|
||||
do_next = 0;
|
||||
for (j=1; j <= NF; j++) {
|
||||
# our stack is deeper, no match
|
||||
if (j > depth_array[i]) {
|
||||
do_next = 1;
|
||||
break;
|
||||
}
|
||||
ignore = ignore_array[i,j];
|
||||
# we found the end of the stack we have a match
|
||||
if (ignore == "**") {
|
||||
match_found = 1;
|
||||
break;
|
||||
}
|
||||
# our stack is mismatched, no match
|
||||
if ((ignore != "*") &&
|
||||
(ignore != $j)) {
|
||||
do_next = 1;
|
||||
break;
|
||||
}
|
||||
}
|
||||
# we ve matched to the end of the stack
|
||||
if (match_found == 1) {
|
||||
break;
|
||||
}
|
||||
# we haven t found a mismatch, make sure the
|
||||
# stack depth is long enough.
|
||||
if (do_next != 1) {
|
||||
if (NF == depth_array[i]) {
|
||||
match_found=1;
|
||||
break;
|
||||
}
|
||||
}
|
||||
}
|
||||
if (match_found == 1) {
|
||||
if (bugnum_array[i] != "") {
|
||||
print "IGNORED STACK (" bugnum_array[i] "): " $0;
|
||||
} else {
|
||||
print "IGNORED STACK: " $0;
|
||||
}
|
||||
} else {
|
||||
print "NEW STACK: " $0;
|
||||
new=1;
|
||||
}
|
||||
}
|
||||
END {
|
||||
exit new;
|
||||
}'
|
||||
ret=$?
|
||||
return $ret
|
||||
}
|
||||
|
||||
############################### parse_log ##############################
|
||||
@@ -700,6 +694,7 @@ cnt_total()
|
||||
echo ""
|
||||
echo "TinderboxPrint:${OPT} Lk bytes: ${tbytes}"
|
||||
echo "TinderboxPrint:${OPT} Lk blocks: ${tblocks}"
|
||||
echo "TinderboxPrint:${OPT} # of runs: ${truns}"
|
||||
echo ""
|
||||
}
|
||||
|
||||
@@ -723,3 +718,4 @@ run_ocsp
|
||||
|
||||
cnt_total
|
||||
memleak_cleanup
|
||||
|
||||
|
||||
Reference in New Issue
Block a user